Upgrading from basic keyword search to something smarter used to mean hiring an ML team. Modern AI changes that – wrap your existing search in an agent and get better results without expensive infrastructure. Here's what we learned experimenting with agent-wrapped product search.
Roaming RAG offers a fresh take on Retrieval-Augmented Generation, letting LLMs navigate well-structured documents like a human—exploring outlines and diving into sections to find answers. Forget complex retrieval setups and vector databases; this streamlined approach delivers rich context and reliable answers with less hassle. It’s perfect for structured content like technical manuals, product…
Learn how to combine the best of keyword and semantic search using SPLADE - a powerful technique that delivers more accurate, transparent, and efficient search results. This practical guide shows you how to implement SPLADE in Elasticsearch to dramatically improve your search capabilities.
Anthropic SKILLs demonstrates the Red Riding Hood Principle perfectly—by leveraging familiar filesystem metaphors that align with Claude's training, SKILLs enable more reliable and intuitive agent behavior. Learn how to apply this principle when building your own AI agents.
Transform your AI applications by thinking like an AI intern on their first day. This post reveals how empathy-driven context engineering leads to better LLM performance, covering everything from handling the intern's ADHD-like attention patterns to providing proper tools and clear instructions. Learn practical strategies for building AI agents that actually understand their tasks and deliver…
Unlock the full potential of AI assistants by mastering artifacts - a game-changing approach that transforms chatbots into powerful collaborative tools. Discover how to build assistants that don't just talk, but actually help you get work done by creating, editing and organizing content alongside you. Complete with practical implementation tips and working demo.
Discover how to build reliable LLM applications by applying the E-Myth's systematic approach to automation. Learn why starting with human processes and incrementally automating tasks leads to more robust LLM applications. Perfect for startup founders and teams struggling to harness the power of LLMs effectively.
Transform your e-commerce search from basic keyword matching to conversational AI—one step at a time. Learn how to incrementally adopt AI without overhauling your existing infrastructure, starting with simple query suggestions and building up to a full conversational assistant. No search experts required, just a thin AI layer that makes your current search engine smarter.
I had my Cursor agent watch a YouTube tutorial on AI video editing and replicate the process – packaged as an agent skill – in about five minutes. One bug fix later, it was editing videos on its own. Including this one!
I built a Cursor command that creates hierarchical explanations and posts them to GitHub issues. But the real story here isn't the tool—it's that I'm literally programming in English. The markdown file is the code, and the LLM agent is its runtime.
I vibe-coded a CLI tool that summarizes YouTube videos, then used it to summarize the video of me making the tool. But the real magic? The 'Recipe' pattern – a one-off, reusable doc for automating repetitive coding tasks. It's like Cursor rules, but you only eat the cookie once.
Natural language specs are inherently ambiguous, and AI agents lack the nuanced context that human developers have. This leads to codebases becoming patchwork quilts and product decisions getting trampled over. I'll explore why current spec-driven development fails at scale and how we might build conversational, living specifications that evolve with our codebases instead of being thrown away.
Turn your LLM into a precision instrument for classification – no fine-tuning required. This post shows how to go beyond simple 'yes/no' answers and unlock soft classification using logprobs. You'll learn how to extract class probabilities in a single call, tune your classifier for optimal performance, and make your LLM behave more like a proper ML model. Perfect for anyone building smarter, more…
Each era of AI product development has absorbed and buried the last. A practitioner's account of five eras – from prompt engineering to agentic runtimes – and what the trajectory tells you about what to build next.
How a Nashville software agency built a dark factory – an AI dev pipeline where 70% of pull requests are approved without a human. Two principles make their software factory actually work.
The next era of AI products will not be built around individual chatbots or coding assistants. It will be built around agents that carry context, tools, skills, and interfaces across every part of digital work. To get there, we need to stop treating the 'agent harness' as a coding-product wrapper and start treating the agent as the product runtime itself.
From silly cat costumes to world-changing innovations, OpenAI's latest release marks the beginning of something extraordinary. The fascinating world of visual reasoning is emerging, where AI models will soon think in pictures and solve complex spatial puzzles, transforming how machines understand and interact with the physical world.